WMT2017 Czech-English machine translation challenge for news

Translate news articles from Czech into English. [ver. 1.0.0]

# submitter when ver. description dev-0 GLEU dev-0 BLEU test-A GLEU test-A BLEU
104 [anonymized] 2021-02-20 14:54 1.0.0 test for ev fairseq m2m-100 just-inference 0.0179 0.0000 0.0292 0.0075
155 [anonymized] 2021-02-18 18:35 1.0.0 files + solution + out 0.0179 0.0000 N/A N/A
120 [anonymized] 2021-02-05 09:22 1.0.0 Add results. pytorch-nn gru 0.0183 0.0000 0.0182 0.0004
137 [anonymized] 2021-02-03 20:59 1.0.0 cz-en 0.0277 0.0016 0.0288 0.0000
136 [anonymized] 2021-02-03 17:38 1.0.0 gru_cz pytorch-nn gru 0.0175 0.0007 0.0170 0.0000
135 [anonymized] 2021-01-30 08:48 1.0.0 Add retrained model. pytorch-nn gru N/A N/A 0.0204 0.0000
134 [anonymized] 2021-01-29 08:34 1.0.0 Add result. N/A N/A 0.0186 0.0000
118 [anonymized] 2021-01-27 04:55 1.0.0 v2.1 lstm pytorch-nn 0.0197 0.0007 0.0192 0.0009
154 [anonymized] 2021-01-27 03:00 1.0.0 lstm lstm pytorch-nn N/A N/A N/A N/A
133 [anonymized] 2021-01-27 01:53 1.0.0 v1.2 0.0100 0.0000 0.0095 0.0000
123 [anonymized] 2021-01-26 01:34 1.0.0 v1.1 0.0112 0.0000 0.0115 0.0000
72 [anonymized] 2020-01-29 07:51 1.0.0 Moses czech to english with 10 corpus sizes trainsize=10000 moses N/A N/A 0.1206 0.0565
71 [anonymized] 2020-01-29 07:51 1.0.0 Moses czech to english with 10 corpus sizes trainsize=1000 moses N/A N/A 0.1206 0.0565
67 [anonymized] 2020-01-29 07:51 1.0.0 Moses czech to english with 10 corpus sizes trainsize=50000 moses N/A N/A 0.1449 0.0822
64 [anonymized] 2020-01-29 07:51 1.0.0 Moses czech to english with 10 corpus sizes trainsize=100000 moses N/A N/A 0.1544 0.0943
60 [anonymized] 2020-01-29 07:51 1.0.0 Moses czech to english with 10 corpus sizes trainsize=198842 moses N/A N/A 0.1639 0.1044
59 [anonymized] 2020-01-29 07:51 1.0.0 Moses czech to english with 10 corpus sizes trainsize=200000 moses N/A N/A 0.1639 0.1045
58 [anonymized] 2020-01-29 07:51 1.0.0 Moses czech to english with 10 corpus sizes trainsize=500000 moses N/A N/A 0.1647 0.1057
57 [anonymized] 2020-01-29 07:51 1.0.0 Moses czech to english with 10 corpus sizes trainsize=400000 moses N/A N/A 0.1647 0.1057
56 [anonymized] 2020-01-29 07:51 1.0.0 Moses czech to english with 10 corpus sizes trainsize=300000 moses N/A N/A 0.1647 0.1057
55 [anonymized] 2020-01-29 07:51 1.0.0 Moses czech to english with 10 corpus sizes trainsize=250000 moses N/A N/A 0.1647 0.1057
48 [anonymized] 2020-01-27 09:33 1.0.0 TAU_28 Marian s2s, tokenizer, truecaser marian 0.2100 0.1592 0.1823 0.1298
22 [anonymized] 2019-12-30 15:02 1.0.0 Runed an available model marian 0.2985 0.2569 0.2918 0.2546
63 [anonymized] 2019-11-20 09:44 1.0.0 Add dev-0 results moses 0.1697 0.1039 0.1576 0.0958
62 [anonymized] 2019-11-13 23:05 1.0.0 moses fixed solution moses 0.2541 0.2042 0.1576 0.0958
85 [anonymized] 2019-11-12 22:51 1.0.0 simple phrase-based system with self-made GROW-DIAG-FINAL, phrase extraction algorithm (Philipp Koehn) and very, very simple decoder self-made 0.0698 0.0173 0.0657 0.0159
115 [anonymized] 2019-11-06 09:24 1.0.0 basic solution (update missing code chunk) self-made word-level 0.0274 0.0030 0.0261 0.0032
92 [anonymized] 2019-11-06 08:25 1.0.0 Solution of IBM methode1 self-made word-level 0.0641 0.0115 0.0452 0.0141
91 [anonymized] 2019-11-06 08:23 1.0.0 Test solution IBM model1 with test-A 0.0641 0.0115 0.0452 0.0141
153 [anonymized] 2019-11-06 08:20 1.0.0 Test solution IBM model1 0.0641 0.0115 N/A N/A
114 [anonymized] 2019-11-05 23:00 1.0.0 basic solution self-made word-level 0.0274 0.0030 0.0261 0.0032
94 [anonymized] 2019-11-05 19:42 1.0.0 300 lines 30 iterations self-made java word-level 0.0464 0.0177 0.0397 0.0134
68 Artur Nowakowski 2019-11-04 15:41 1.0.0 GIZA++ IBM Model 1 word-level 0.1727 0.0911 0.1473 0.0700
80 [anonymized] 2019-11-02 09:56 1.0.0 now its ok, but previous was strange self-made word-level 0.0800 0.0248 0.0626 0.0189
152 [anonymized] 2019-11-02 09:49 1.0.0 my local geval thinks that out files have too many lines N/A N/A N/A N/A
97 [anonymized] 2019-11-02 07:06 1.0.0 Second attempt self-made word-level 0.0487 0.0150 0.0415 0.0114
95 [anonymized] 2019-11-01 14:06 1.0.0 Another try with IBM MODEL 1 self-made word-level 0.0760 0.0151 0.0685 0.0127
70 [anonymized] 2019-10-31 13:13 1.0.0 use giza++ for word alignment word-level 0.1441 0.0730 0.1273 0.0614
151 [anonymized] 2019-10-29 22:48 1.0.0 final stupid solution Y stupid 0.0219 0.0018 N/A N/A
150 [anonymized] 2019-10-29 22:25 1.0.0 final stupid solution 0.0188 0.0000 N/A N/A
102 [anonymized] 2019-10-29 17:59 1.0.0 First attempt self-made word-level 0.0540 0.0139 0.0456 0.0099
27 [anonymized] 2019-10-29 08:07 1.0.0 Moses with MERT tuning moses 0.2541 0.2041 0.2071 0.1542
117 [anonymized] 2019-10-23 12:01 1.0.0 add main.py N/A N/A 0.0490 0.0031
116 [anonymized] 2019-10-23 08:54 1.0.0 stupid solution stupid N/A N/A 0.0490 0.0031
76 [anonymized] 2019-10-23 08:19 1.0.0 Solution stupid N/A N/A 0.0846 0.0329
26 Artur Nowakowski 2019-10-23 08:13 1.0.0 Moses with MERT tuning moses 0.2443 0.1929 0.2191 0.1675
81 [anonymized] 2019-10-22 22:25 1.0.0 stupid solution stupid 0.0440 0.0080 0.0481 0.0184
105 [anonymized] 2019-10-22 20:42 1.0.0 example stupid stupid 0.0489 0.0075 0.0423 0.0068
119 [anonymized] 2019-10-22 18:22 1.0.0 300 most common tetragrams stupid java 0.0204 0.0010 0.0190 0.0008
112 [anonymized] 2019-10-22 15:03 1.0.0 sialala one more stupid solution stupid 0.0351 0.0032 0.0352 0.0038
132 [anonymized] 2019-10-22 14:29 1.0.0 sialala stupid solution 0.0424 0.0017 0.0389 0.0000
106 [anonymized] 2019-10-22 11:01 1.0.0 czech - eng translation stupid 0.0446 0.0074 0.0364 0.0058
90 [anonymized] 2019-10-22 09:38 1.0.0 cp in.tsv out.tsv stupid 0.0464 0.0175 0.0420 0.0151
93 [anonymized] 2019-10-21 21:13 1.0.0 tlumaczenie stupid N/A N/A 0.0603 0.0141
113 [anonymized] 2019-10-21 10:27 1.0.0 Stupid solution stupid 0.0308 0.0041 0.0258 0.0037
110 [anonymized] 2019-10-20 19:52 1.0.0 first solution stupid 0.0302 0.0081 0.0285 0.0040
121 [anonymized] 2019-10-20 18:23 1.0.0 stupid solution stupid 0.0166 0.0006 0.0145 0.0004
131 [anonymized] 2019-10-19 22:56 1.0.0 stuuupid plus common words 0.0395 0.0015 0.0364 0.0000
130 [anonymized] 2019-10-19 20:58 1.0.0 no potato 0.0372 0.0014 0.0364 0.0000
129 [anonymized] 2019-10-19 20:48 1.0.0 stuuupid solution 0.0372 0.0014 0.0343 0.0000
79 [anonymized] 2019-10-19 17:38 1.0.0 My stupid solution stupid 0.0802 0.0255 0.0724 0.0217
82 [anonymized] 2019-10-18 12:10 1.0.0 TAU2019-001 stupid 0.0548 0.0209 0.0474 0.0178
107 [anonymized] 2019-10-17 12:40 1.0.0 Simple task with non zero BLEU stupid 0.0407 0.0072 0.0327 0.0056
108 [anonymized] 2019-10-16 10:49 1.0.0 stupid best 4-gram sentence stupid 0.0365 0.0095 0.0327 0.0056
149 [anonymized] 2019-10-16 10:44 1.0.0 4-gram sentences 0.0365 0.0095 N/A N/A
2 [anonymized] 2019-10-15 21:03 1.0.0 lab1 - substitute existing 0.0639 0.0198 0.3373 0.3095
1 Artur Nowakowski 2019-10-15 13:19 1.0.0 Substitute solution existing 0.0631 0.0196 0.3373 0.3095
21 [anonymized] 2019-10-14 12:08 1.0.0 Regex solution 0.0691 0.0292 0.3067 0.2689
28 [anonymized] 2019-02-08 09:05 1.0.0 TAU-2018-020 marian 0.0507 0.0273 0.1886 0.1435
128 [anonymized] 2019-02-08 09:03 1.0.0 TAU-2018-016 neural-network word-level 0.0225 0.0009 0.0221 0.0000
49 [anonymized] 2019-02-07 22:09 1.0.0 a-an try v4 0.1944 0.1452 0.1696 0.1212
53 [anonymized] 2019-02-07 21:50 1.0.0 a-an v3 0.1936 0.1444 0.1689 0.1205
51 [anonymized] 2019-02-07 21:47 1.0.0 a-an v2 0.1944 0.1452 0.1696 0.1212
52 [anonymized] 2019-02-07 21:26 1.0.0 try X 0.1937 0.1446 0.1689 0.1206
41 [anonymized] 2019-02-05 20:27 1.0.0 Cheap fixed translations 0.2142 0.1689 0.1880 0.1425
42 [anonymized] 2019-02-05 20:17 1.0.0 Cheap fixed copying and translation 0.2142 0.1689 0.1880 0.1425
40 [anonymized] 2019-02-05 19:56 1.0.0 Add -year-old for test-A as well 0.2142 0.1689 0.1879 0.1426
43 [anonymized] 2019-02-05 19:49 1.0.0 Add manual female surnames copying 0.2140 0.1687 0.1875 0.1419
29 [anonymized] 2019-02-05 19:09 1.0.0 Post process Marian TAU-2018-020 marian 0.2142 0.1689 0.1880 0.1426
50 [anonymized] 2019-02-04 22:46 1.0.0 a / an try marian 0.1944 0.1452 0.1696 0.1212
34 [anonymized] 2019-02-04 22:20 1.0.0 Fix often faulty Rio Summer Olympics translation 0.2142 0.1689 0.1879 0.1426
30 [anonymized] 2019-02-04 22:14 1.0.0 Add one-off fixes for weird translations 0.2142 0.1689 0.1880 0.1426
33 [anonymized] 2019-02-04 21:24 1.0.0 Fix dollar <space $ space> 0.2141 0.1688 0.1879 0.1426
32 [anonymized] 2019-02-04 20:08 1.0.0 More interpunction removed 0.2141 0.1688 0.1879 0.1426
31 [anonymized] 2019-02-03 21:13 1.0.0 Replace " <interpunction>" with "<interpunction>" 0.2141 0.1688 0.1879 0.1426
39 [anonymized] 2019-02-03 19:16 1.0.0 marian baseline epoch 5 0.2141 0.1687 0.1879 0.1426
38 [anonymized] 2019-01-30 10:20 1.0.0 wfwbfei 0.2142 0.1688 0.1879 0.1426
37 [anonymized] 2019-01-30 10:15 1.0.0 3 0.2142 0.1688 0.1879 0.1426
148 [anonymized] 2019-01-30 10:04 1.0.0 next N/A N/A N/A N/A
147 [anonymized] 2019-01-30 10:03 1.0.0 2 N/A N/A N/A N/A
36 [anonymized] 2019-01-30 09:52 1.0.0 full files marian 0.2141 0.1688 0.1879 0.1426
146 [anonymized] 2019-01-30 09:46 1.0.0 marian N/A N/A N/A N/A
145 [anonymized] 2019-01-30 09:44 1.0.0 my try N/A N/A N/A N/A
35 p/tlen 2019-01-27 16:13 1.0.0 Marian 5 epochs marian 0.2141 0.1688 0.1879 0.1426
144 [anonymized] 2019-01-23 07:56 1.0.0 beginning neural-network word-level N/A N/A N/A N/A
73 [anonymized] 2019-01-09 10:30 1.0.0 ex 03 existing 0.1370 0.0592 0.1252 0.0509
83 [anonymized] 2019-01-09 09:20 1.0.0 merge and ex01 simple non-zero 0.0504 0.0198 0.0437 0.0163
143 [anonymized] 2019-01-02 08:40 1.0.0 neural network translate by words - Dawid Kubicki neural-network word-level 0.0548 0.0127 N/A N/A
54 [anonymized] 2018-11-07 11:15 1.0.0 Moses europarl tuned 0.1887 0.1295 0.1671 0.1079
84 [anonymized] 2018-11-07 09:39 1.0.0 GEVAL from IBM Model 1 self-made 0.0450 0.0000 0.0805 0.0162
142 [anonymized] 2018-11-07 09:08 1.0.0 tau-003-improve 0.0586 0.0163 N/A N/A
66 [anonymized] 2018-11-07 00:04 1.0.0 moses moses 0.1658 0.1002 0.1505 0.0871
69 [anonymized] 2018-11-06 20:13 1.0.0 Self made fast align self-made 0.1619 0.0795 0.1473 0.0662
5 [anonymized] 2018-11-06 20:12 1.0.0 translation IBM MODEL 1 files self-made existing 0.0450 0.0000 0.3324 0.3039
61 [anonymized] 2018-11-06 18:31 1.0.0 Moses europarl no-tune moses 0.1864 0.1260 0.1632 0.1030
4 [anonymized] 2018-11-06 18:26 1.0.0 solution existing 0.0586 0.0163 0.3324 0.3039
47 [anonymized] 2018-10-31 18:58 1.0.0 moses solution moses 0.2140 0.1525 0.1911 0.1322
46 [anonymized] 2018-10-31 18:46 1.0.0 Moses 0.2140 0.1525 0.1911 0.1322
44 [anonymized] 2018-10-31 10:58 1.0.0 moses solution moses 0.2221 0.1640 0.1979 0.1393
11 [anonymized] 2018-10-30 11:21 1.0.0 improve online-B existing 0.0645 0.0258 0.3079 0.2704
3 [anonymized] 2018-10-29 15:56 1.0.0 Simple regex improvement TAU-2018-003.py existing 0.0450 0.0000 0.3324 0.3039
45 [anonymized] 2018-10-27 12:09 1.0.0 Moses v1 moses 0.2032 0.1428 0.1911 0.1323
20 [anonymized] 2018-10-24 09:07 1.0.0 online-B existing N/A N/A 0.3067 0.2689
24 [anonymized] 2018-10-24 09:03 1.0.0 ex02 existing N/A N/A 0.2862 0.2465
19 [anonymized] 2018-10-24 08:22 1.0.0 TAU 002 existing 0.0586 0.0163 0.3067 0.2689
18 [anonymized] 2018-10-24 08:16 1.0.0 002 existing 0.0645 0.0258 0.3067 0.2689
17 [anonymized] 2018-10-24 08:15 1.0.0 TAU-2018-002 existing 0.0645 0.0258 0.3067 0.2689
10 [anonymized] 2018-10-24 08:12 1.0.0 uedin dev existing N/A N/A 0.3324 0.3038
9 [anonymized] 2018-10-23 16:14 1.0.0 uedin existing 0.0586 0.0163 0.3324 0.3038
25 [anonymized] 2018-10-23 16:10 1.0.0 PJATK existing 0.0586 0.0163 0.2739 0.2324
16 [anonymized] 2018-10-23 14:25 1.0.0 TAU-2018-002 existing N/A N/A 0.3067 0.2689
23 [anonymized] 2018-10-23 12:56 1.0.0 Online-A existing 0.1259 0.0502 0.2862 0.2465
8 [anonymized] 2018-10-22 12:13 1.0.0 uedin existing 0.0709 0.0272 0.3324 0.3038
15 [anonymized] 2018-10-17 20:00 1.0.0 TAU 002 - OnlineB existing 0.1421 0.1451 0.3067 0.2689
14 [anonymized] 2018-10-17 15:39 1.0.0 online-B solution existing 0.0645 0.0258 0.3067 0.2689
13 [anonymized] 2018-10-17 10:22 1.0.0 zadanie Dawid Kubicki existing N/A N/A 0.3067 0.2689
141 [anonymized] 2018-10-17 10:13 1.0.0 zadanie Dawid Kubicki N/A N/A N/A N/A
12 [anonymized] 2018-10-17 09:27 1.0.0 zadanie existing N/A N/A 0.3067 0.2689
7 [anonymized] 2018-10-17 09:17 1.0.0 rozwiazanie uedin-nmt existing 0.0450 0.0000 0.3324 0.3038
6 [anonymized] 2018-10-17 09:15 1.0.0 Uedin-nmt existing N/A N/A 0.3324 0.3038
77 [anonymized] 2018-10-17 08:32 1.0.0 third commit simple non-zero 0.0709 0.0272 0.0657 0.0238
89 [anonymized] 2018-10-17 08:27 1.0.0 second commit simple non-zero 0.0464 0.0175 0.0420 0.0151
75 [anonymized] 2018-10-17 08:10 1.0.0 Dodane src simple non-zero 0.1259 0.0502 0.1158 0.0434
88 [anonymized] 2018-10-17 08:01 1.0.0 first commit simple non-zero 0.0464 0.0175 0.0420 0.0151
74 [anonymized] 2018-10-16 19:53 1.0.0 Auto word dictionary simple non-zero 0.1259 0.0502 0.1158 0.0434
78 [anonymized] 2018-10-16 16:08 1.0.0 simple solution simple non-zero 0.0645 0.0258 0.0590 0.0223
98 [anonymized] 2018-10-16 14:04 1.0.0 in place replacement 0.0450 0.0000 0.0301 0.0114
101 [anonymized] 2018-10-16 13:55 1.0.0 +Common words -special names 0.0450 0.0000 0.0333 0.0107
86 [anonymized] 2018-10-16 13:36 1.0.0 my solution simple non-zero 0.0586 0.0163 0.0551 0.0157
109 [anonymized] 2018-10-16 11:41 1.0.0 special names without odmiany 0.0450 0.0000 0.0282 0.0048
111 [anonymized] 2018-10-16 11:36 1.0.0 own_names 0.0450 0.0000 0.0271 0.0038
99 [anonymized] 2018-10-16 11:30 1.0.0 three letters acronyms 0.0450 0.0000 0.0338 0.0111
100 [anonymized] 2018-10-16 11:28 1.0.0 acronym 2+ 0.0450 0.0000 0.0335 0.0108
103 [anonymized] 2018-10-16 10:09 1.0.0 +Ancronyms 0.0450 0.0000 0.0310 0.0084
96 [anonymized] 2018-10-16 10:05 1.0.0 numbers + copypaste attack 0.0450 0.0000 0.0352 0.0119
140 [anonymized] 2018-10-15 18:16 1.0.0 ? 0.0464 0.0175 N/A N/A
127 [anonymized] 2018-10-15 18:14 1.0.0 frequency attack using average wordcount N/A N/A 0.0458 0.0000
126 [anonymized] 2018-10-15 18:09 1.0.0 7 words actual file N/A N/A 0.0400 0.0000
125 [anonymized] 2018-10-15 18:08 1.0.0 7 words N/A N/A 0.0342 0.0000
124 [anonymized] 2018-10-15 17:42 1.0.0 frequency 5 word attack N/A N/A 0.0342 0.0000
122 [anonymized] 2018-10-15 16:42 1.0.0 frequency attack N/A N/A 0.0195 0.0000
87 [anonymized] 2018-10-15 16:03 1.0.0 keep testing for too few lines simple non-zero N/A N/A 0.0420 0.0151
139 [anonymized] 2018-10-15 16:02 1.0.0 test submission A N/A N/A N/A N/A
138 [anonymized] 2018-10-15 15:58 1.0.0 test submission N/A N/A N/A N/A
65 [anonymized] 2018-10-14 22:17 1.0.0 First solution simple non-zero 0.1421 0.1451 0.0893 0.0877

Submission graph

Graphs by parameters

trainsize

[direct link]
[direct link]